View Based Methods can achieve Bayes-Optimal 3D Recognition - Computer Science > Computer Vision and Pattern RecognitionReportar como inadecuado




View Based Methods can achieve Bayes-Optimal 3D Recognition - Computer Science > Computer Vision and Pattern Recognition - Descarga este documento en PDF. Documentación en PDF para descargar gratis. Disponible también para leer online.

Abstract: This paper proves that visual object recognition systems using only 2DEuclidean similarity measurements to compare object views against previouslyseen views can achieve the same recognition performance as observers havingaccess to all coordinate information and able of using arbitrary 3D modelsinternally. Furthermore, it demonstrates that such systems do not require moretraining views than Bayes-optimal 3D model-based systems. For building computervision systems, these results imply that using view-based or appearance-basedtechniques with carefully constructed combination of evidence mechanisms maynot be at a disadvantage relative to 3D model-based systems. For computationalapproaches to human vision, they show that it is impossible to distinguishview-based and 3D model-based techniques for 3D object recognition solely bycomparing the performance achievable by human and 3D model-based systems.}



Autor: Thomas M. Breuel

Fuente: https://arxiv.org/







Documentos relacionados